The large-scale use of fossil fuels has resulted in various environmental problems. Accordingly, the development and utilization of renewable and environmentally friendly energy are needed [1-3]. Hydrogen is a promising alternative to traditional fossil fuels because it is a clean energy source with a high heat value [4, 5]. Water splitting inspired by photosynthesis is considered an ideal pathway for the conversion of solar energy and water into hydrogen energy [6-8]. The overall water splitting process comprises two half reactions, namely water oxidation (2H₂O → 4H⁺ + O₂ + 4e^–) and proton reduction (4H⁺ + 4e^– → 2H₂). The water oxidation step has a high activation energy and is regarded as the limiting step of water splitting [9-11]. To improve the efficiency of water oxidation, much research has been focused on exploring water oxidation catalysts (WOCs) [3, 12, 13]. However, the development of efficient WOCs remains challenging. Owing to their low cost and abundance, much recent research effort has focused on WOCs based on first row transitional metals [14-25]. Compared with heterogeneous WOCs, homogeneous WOCs, especially molecular organic complexes, have advantages in structure modification and mechanism research [26]. There are two main categories of homogeneous WOCs, namely inorganic polyoxometallates and organic ligand-based complexes.

Among the homogeneous WOCs, organic complexes containing Co have been widely investigated because they usually show good catalytic activity in light-driven water oxidation. Polycore complexes, such as cubane-like complexes with M₄O₄ (M = transition metal) frames, are attractive because of their biomimetic structures [27-31], which are similar to that of the oxygen-evolving Mn₄CaO₅ cluster in natural photosynthesis system Ⅱ (PSⅡ). For example, a Co₄O₄(Ac)₄(py)₄ cluster and related derivatives were initially regarded as homogeneous WOCs [27, 32]. However, a follow-up study by Nocera et al. [33] suggested that pure Co₄O₄(Ac)₄(py)₄ cluster was inactive. Photochemical and electrochemical experiments confirmed that Co(Ⅱ) impurities were responsible for the apparent catalytic ability [33]. A recent study by Lu et al. [34] showed that two dinuclear cobalt polypyridine complexes containing μ-OH and μ-O₂ core were unable to catalyze water oxidation. Chelating experiments and controlled potential electrolysis measurements attributed the observed catalytic activity of these synthesized samples to Co(Ⅱ) impurities introduced during synthesis.

Sufficient experiments should be carried out to identify true WOCs, especially for Co-based molecular catalysts. Co(Ⅱ) impurities can transform into CoO_x species, which are true WOCs. Therefore, the possibility that the catalytic activity originates from impurities should be ruled out before identifying substrates as true homogeneous WOCs.

Herein, binuclear complex μ-OH, μ-O₂-[{(enN4)₂Co₂}](ClO₄)₃ (1(ClO₄)₃, enN4 = 1, 6-bis(2-pyridyl)-2, 5-diazahex-1, 5-diene) (Fig. 1) was synthesized and used to catalyze water oxidation. Its catalytic ability was investigated using a series of methods, which showed that 1 was inactive as a catalyst for light-driven water oxidation. The apparent activity of the as-synthesized catalyst originated from Co impurities introduced during synthesis. This finding supplements previous reported results and establishes that complexes containing a μ-OH, μ-O₂ dicobalt(Ⅲ) core might be unsuitable as WOCs for light-driven water oxidation. This rule provides guidelines for the future exploration of efficient WOCs.

Purified water (18.2 MΩ·cm) for the preparation of buffer solutions was obtained from a Molecular Lab Water Purifier. All chemical agents were commercially available and used without further purification.

The enN4 ligand was synthesized according to a literature procedure [35]. 2-Pyridinecarboxaldehyde (3.2 g, 30 mmol) and ethylenediamine (0.9 g, 15 mmol) were dissolved in methanol (25 mL). The solution was refluxed for about 7 h, cooled to room temperature, and concentrated by rotary evaporation to obtain a yellow solid. A pure ligand sample was obtained by recrystallization from n-hexane followed by drying in a vacuum oven.

Compound 1 was synthesized according to a literature procedure [36] in the form of 1(ClO₄)₃·2MeCN·H₂O accompanied by one molecule of py-HClO₄. This microcrystalline solid was obtained by adding Co(ClO₄)₂·6H₂O to an acetonitrile solution of 1, 6-bis(2-pyridyl)-2, 5-diazahexa-1, 5-diene and a small quantity of pyridine. The solution became red-brown within minutes, and then changed to dark brown. After stirring for 16 h under air to complete the metal oxidation, black microcrystals were formed. These crystals were collected by filtration and washed with Et₂O several times. Crystals suitable for X-ray diffraction analysis were obtained by slowly evaporating a saturated solution of the black microcrystals in acetonitrile/toluene (3:1, v/v). The addition of a few drops of pyridine aided crystal formation. Elemental analysis of C₃₇H₄₃Cl₄Co₂N₁₁O₂₀ gave the following results: calcd, C 36.38%, H 3.55%, N 12.61%; found, C 35.82%, H 3.45%, N 12.70%.

Cyclic voltammetry (CV) was recorded on a CHI660D electrochemical analyzer. Glassy carbon (GC; 0.071 cm²), Ag/AgCl (3.5 mol/L KCl), and Pt wire were used as the working, reference, and auxiliary electrodes, respectively. CV tests were carried out in buffer solutions at pH = 9.0 and room temperature using a scan rate of 100 mV/s. For controlled potential electrolysis experiments, indium-tin oxide (ITO) glass slides (2.0 cm × 1.0 cm, 1.0 cm² immersed in electrolyte), Ag/AgCl, and Pt wire were used as the working, reference, and auxiliary electrodes, respectively. ITO glass slides were cleaned by sonication in acetone, ethanol, and ultrapure water for 30 min, successively.

Photocatalytic water oxidation was performed as follows. Catalyst was added to the reaction solution (80 mmol/L borate buffer solution) containing [Ru(bpy)₃]Cl₂ (bpy = 2, 2′-bipyridine) (1 mmol/L) and Na₂S₂O₈ (10 mmol/L). The above solution was deaerated by purging with Ar gas for 10 min in a flask (total volume = 28 mL) and then sealed with a rubber septum (reaction solution volume = 15 mL). The reaction was initiated by irradiating the solution using light-emitting diodes (LEDs) as the light source (light intensity = 15.8 mW, beam diameter = 2 cm) through a transmitting glass filter (λ ≥ 420 nm) at ambient temperature. After each irradiation period, Ar gas (150 μL) was injected into the flask and the same volume of gas was withdrawn from the flask headspace using an SGE gas-tight syringe and analyzed by gas chromatography. The oxygen in the sampled gas was separated by passing through a packed molecular sieve 5A column (2 m × 3 mm) using Ar as the carrier gas and quantified using a thermal conductivity detector (TCD, Shimadzu GC-9A). The total amount of evolved oxygen was calculated from the oxygen concentration in the headspace. Contamination of the headspace with air was corrected by measuring the nitrogen present in the headspace from the nitrogen peak area in the gas chromatography traces.

Single-crystal X-ray diffraction analysis was performed in a cooled nitrogen gas stream at 100 K on an Agilent Technologies supernova diffractometer equipped with a charge-coupled device bidimensional detector. UV-Vis absorption spectra were recorded on a TU-1810 spectrophotometer (Beijing Purkinje General Instrument Co., Ltd.) equipped with a photomultiplier tube detector.

Fig. 2 and Fig. 3 show the X-ray single crystal structure of 1. The two Co^Ⅲ ions were linked by coordination to two 1, 6-bis(2-pyridyl-2, 5-diazaocta-2, 6-diene ligands, one μ-OH ligand, and one μ-O₂ ligand (Table 1-3). Complex 1(ClO₄)₃ was further characterized by Fourier transform infrared (FTIR) spectroscopy. Notable IR bands were observed at 883, 772, and 651 cm^–1 in the lower energy region (Fig. 4). The bands at 883 cm^–1 were assigned to v(O–O) [36].

Together, the as-synthesized catalyst, sacrificial electron acceptor Na₂S₂O₈, and photosensitizer [Ru(bpy)₃]Cl₂ constituted the artificial photosynthesis oxygen evolution system. When 10 mmol/L Na₂S₂O₈ and 1 mmol/L [Ru(bpy)₃]Cl₂ were used, an optimal oxygen yield of 46.7% was achieved (Fig. 5(a)). To investigate the true catalytic ability of 1, the light-driven water oxidation reaction was repeated under similar conditions except with bpy (0.2 or 0.5 equiv.) added as a chelating agent. When 2 μmol/L bpy was added to chelate the as-synthesized catalyst (10 μmol/L), the oxygen yield was reduced to 9.6%. A further increase in the bpy concentration to 10 μmol/L led to complete deactivation of the catalytic system (Fig. 5(b)). Previous studies by Nocera et al. [33] and Lu et al. [34] showed that this phenomenon was due to Co(Ⅱ) impurities residing in the as-synthesized catalyst. Co(Ⅱ) impurities in the as-synthesized catalyst can transform into CoO_x, which is a true WOC. The added bpy chelates Co impurities to produce the catalytically inactive Co-bipyridine complex. However, the addition of chelating agent does not affect the oxygen evolution system catalyzed by genuine homogeneous catalysts [31].

Fig. 6(a) shows UV-Vis absorption spectra of the as-synthesized catalyst in pH = 9.0 borate buffer solution with and without bpy, which are identical, indicating that bpy did not destroy the structure of 1. The UV-Vis absorption spectrum of the as-synthesized catalyst retained its characteristic peaks before and after the addition of Na₂S₂O₈, indicating that Na₂S₂O₈ did not destroy the structure of 1 without irradiation.

The catalytic ability of 1 was further investigated using electrochemical experiments. First, CV measurements of the as-synthesized catalyst were performed in pH = 9.0 borate buffer solution. As shown in Fig. 7(a), the borate buffer solution produced a negligible catalytic current. However, an increased catalytic current was observed during repeated CV scanning of the as-synthesized catalyst, which is a typical characteristic of the electrodeposition of heterogeneous active species as true WOCs on the surface of a GC electrode. The electrochemical behavior of the as-synthesized catalyst was similar to that of free Co²⁺ ions (Fig. 7(b)). After ten cyclic scans, the used GC electrode was rinsed with ultrapure water, but not polished, and retested by CV in fresh pH = 9.0 borate buffer solution without addition of the as-synthesized catalyst. As expected, a significant catalytic current was produced (Fig. 8), demonstrating the presence of an active heterogeneous Co species on the surface of the GC electrode.

Co element in the active species may originate from impurities, or may have been released from μ-OH, μ-O₂-[{(enN4)₂Co₂}](ClO₄)₃ during the CV test. To identify the source of Co element, CV experiments of the as-synthesized catalyst were conducted in the presence of bpy. Fig. 9 illustrates the catalytic current of as-synthesized catalyst (0.5 mmol/L) in borate buffer solution containing different concentrations of bpy. The catalytic currents at 1.51 V vs. NHE decreased with increasing bpy concentration. The Co-bipyridine complex was inactive for electrochemical water oxidation. CV tests of borate buffer solution containing as-synthesized catalyst sample (0.5 mmol/L) and bpy (0.25 mmol/L or 0.5 mmol/L) gave a negligible catalytic current compared with the background. UV-Vis absorption spectra verified that bpy did not destroy the chemical structure of 1 and that it only chelated Co impurities and not 1. Therefore, 1 was inactive as a catalyst for electrocatalytic water oxidation.

Controlled potential electrolysis (CPE) experiments were used to examine the catalytic ability of 1. As mentioned above, ITO was used as the working electrode instead of GC in all CPE tests. During CPE of the as-synthesized catalyst (0.5 mmol/L) in borate buffer at pH = 9.0, the catalytic current increased after the start of CPE, and then reached a maximum of 0.55 mA/cm² after 4 h (Fig. 10). ITO used for the above CPE experiment (ITO-1) was washed with ultrapure water and retested in CPE using fresh borate buffer solution. As expected, the used ITO electrode exhibited a high catalytic current compared with the background in fresh borate buffer at pH = 9.0. According to previous reports [34, 37], the catalytic activity of used ITO is due to residual active species on the electrode surface.

The CPE experiment of as-synthesized catalyst (0.5 mM) in borate buffer solution containing 0.25 mmol/L bpy at 1.31 V vs. NHE is shown in Fig. 11. ITO used in this experiment is denoted as ITO-2. In the presence of bpy, a completely different catalytic current was produced. The catalytic current of borate buffer solution containing as-synthesized catalyst (0.5 mmol/L) and bpy (0.25 mmol/L) was negligible compared with that of the borate buffer solution without bpy addition. As the impurities were chelated by bpy, the negligible catalytic current (compared with the catalytic current of the as-synthesized catalyst sample, black line in Fig. 11) was attributed to 1 alone.

A yellow residue was present on the surface of ITO-1 (Fig. 12). This residue was also observed in the scanning electron microscopy (SEM) image of ITO-1 (Fig. 13(a)). However, the SEM image of ITO-2 (Fig. 13(b)) was no different from that of the clean ITO electrode (Fig. 13(c)). Furthermore, energy dispersive spectroscopy (EDS) analysis of the ITO electrodes was in good agreement with the SEM results. Co element was only detected on the surface of ITO-1 (Fig. 14(a)), while no Co was present on the surface of ITO-2 (Fig. 14(b)), which was similar to the clean ITO electrode (Fig. 14(c)). Therefore, pure 1 was stable during the CPE test and did not decompose into heterogeneous Co species. The Co species responsible for water oxidation came from Co impurities. The above CPE experiments demonstrated the following: First, the observed catalytic capability of the as-synthesized catalyst was derived from Co impurities; second, bpy stabilized the Co impurities by generating a stable species that was inactive for water oxidation; and third, the synthesized sample lost activity when the Co impurities were chelated by bpy. Therefore, pure 1 was unable to catalyze water oxidation.

The process of catalyst deposition onto the ITO surface was further investigated by XPS. Fig. 15(a) shows XPS analysis of Co element on the ITO electrode obtained from the CPE of borate buffer solution (pH = 9.0) containing 0.5 mmol/L Co(ClO₄)₂·6H₂O (ITO-3), which generated active Co species for water oxidation [34]. Two peaks with binding energies of 780.5 and 795.9 eV were assigned to Co 2p_3/2 and Co 2p_1/2, respectively. XPS analysis of Co element on ITO-1 showed two peaks at 780.6 and 796.0 eV (Fig. 15(b)), which was very similar to that of ITO-3 (Fig. 15(c)), indicating that Co impurities and free Co²⁺ had the same deposition behavior. Therefore, the Co impurities present in the as-synthesized catalyst had a valency of +2.

Herein, we investigated the catalytic ability of a dinuclear cobalt polypyridine complex containing μ-OH and μ-O₂ ligands in the core structure. This complex was confirmed as inactive for water oxidation using photochemical and electrochemical measurements. In combination with previous reports by others, this work establishes that dinuclear cobalt polypyridine complexes containing μ-OH and μ-O₂ structure do not catalyze water oxidation, with true WOCs requiring another specific chemical structure. This work provides new insight toward the exploration of new catalysts for water oxidation.